image.png

Walmart Black Friday Purchases¶

The project aims to analyze the customer purchase behavior (specifically, purchase amount) against the customer’s gender and the various other factors to help the business make better decisions. They want to understand if the spending habits differ between male and female customers: Do women spend more on Black Friday than men?

Bottom Line Upfront (BLUF):¶

Through the use of confidence intervals (CI) and bootstrap methods, we identified key trends in spending patterns.

  • Males spend significantly more than females, with average male purchases around \$925,408, compared to \$712,217 for females.
  • Single men tend to spend more than married men, suggesting that marital status influences spending behavior, particularly for men.
  • Customers aged 25-40 represent the largest proportion of high spenders (over 80%).
  • City B drives the most overall sales volume, but City C generates higher average purchases per customer.
  • Product categories 6 and 15 are the most expensive, while categories 1, 5, 8, and 11 have the highest purchase frequency.

Recommendations:

  • Target Male Shoppers: Focus on high-value product promotions for males, as they spend significantly more than females.
  • Single Men Focus: Single men spend more than married men, so consider tailored offers for this group.
  • Age-Based Strategy: Promote premium products to older age groups (51+), who tend to make higher purchases.
  • City-Specific Offers: Target City C with luxury items due to higher average spending, and use discounts to increase transaction sizes in City B.
  • Product Categories: Prioritize promotions for high-value categories (6 and 15) and boost sales in frequent purchase categories (1, 5, 8, and 11) with bundled deals.

Library¶

In [1]:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
sns.set(color_codes = True)
import scipy.stats as stats
from scipy.stats import norm
import warnings
warnings.filterwarnings("ignore")

EDA¶

Data Overview¶

In [2]:
#load data
df = pd.read_csv('walmart_data.csv')
df.head()
Out[2]:
User_ID Product_ID Gender Age Occupation City_Category Stay_In_Current_City_Years Marital_Status Product_Category Purchase
0 1000001 P00069042 F 0-17 10 A 2 0 3 8370
1 1000001 P00248942 F 0-17 10 A 2 0 1 15200
2 1000001 P00087842 F 0-17 10 A 2 0 12 1422
3 1000001 P00085442 F 0-17 10 A 2 0 12 1057
4 1000002 P00285442 M 55+ 16 C 4+ 0 8 7969
In [3]:
df.tail()
Out[3]:
User_ID Product_ID Gender Age Occupation City_Category Stay_In_Current_City_Years Marital_Status Product_Category Purchase
550063 1006033 P00372445 M 51-55 13 B 1 1 20 368
550064 1006035 P00375436 F 26-35 1 C 3 0 20 371
550065 1006036 P00375436 F 26-35 15 B 4+ 1 20 137
550066 1006038 P00375436 F 55+ 1 C 2 0 20 365
550067 1006039 P00371644 F 46-50 0 B 4+ 1 20 490
In [4]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 550068 entries, 0 to 550067
Data columns (total 10 columns):
 #   Column                      Non-Null Count   Dtype 
---  ------                      --------------   ----- 
 0   User_ID                     550068 non-null  int64 
 1   Product_ID                  550068 non-null  object
 2   Gender                      550068 non-null  object
 3   Age                         550068 non-null  object
 4   Occupation                  550068 non-null  int64 
 5   City_Category               550068 non-null  object
 6   Stay_In_Current_City_Years  550068 non-null  object
 7   Marital_Status              550068 non-null  int64 
 8   Product_Category            550068 non-null  int64 
 9   Purchase                    550068 non-null  int64 
dtypes: int64(5), object(5)
memory usage: 42.0+ MB
In [5]:
df.shape
Out[5]:
(550068, 10)
In [6]:
df.columns
Out[6]:
Index(['User_ID', 'Product_ID', 'Gender', 'Age', 'Occupation', 'City_Category',
       'Stay_In_Current_City_Years', 'Marital_Status', 'Product_Category',
       'Purchase'],
      dtype='object')
In [7]:
df.dtypes
Out[7]:
User_ID                        int64
Product_ID                    object
Gender                        object
Age                           object
Occupation                     int64
City_Category                 object
Stay_In_Current_City_Years    object
Marital_Status                 int64
Product_Category               int64
Purchase                       int64
dtype: object

Finding Unique Values and Count Them¶

We will find out the unique values in each column. This will help us understand how varied the data is.

In [8]:
def print_unique_values(df):
  for column in df.columns:
    unique_values = df[column].unique()
    print(f"\nUnique Values of {column}:", unique_values)

print_unique_values(df)

Unique Values of User_ID: [1000001 1000002 1000003 ... 1004113 1005391 1001529]

Unique Values of Product_ID: ['P00069042' 'P00248942' 'P00087842' ... 'P00370293' 'P00371644'
 'P00370853']

Unique Values of Gender: ['F' 'M']

Unique Values of Age: ['0-17' '55+' '26-35' '46-50' '51-55' '36-45' '18-25']

Unique Values of Occupation: [10 16 15  7 20  9  1 12 17  0  3  4 11  8 19  2 18  5 14 13  6]

Unique Values of City_Category: ['A' 'C' 'B']

Unique Values of Stay_In_Current_City_Years: ['2' '4+' '3' '1' '0']

Unique Values of Marital_Status: [0 1]

Unique Values of Product_Category: [ 3  1 12  8  5  4  2  6 14 11 13 15  7 16 18 10 17  9 20 19]

Unique Values of Purchase: [ 8370 15200  1422 ...   135   123   613]
In [9]:
def print_nunique_values(df):
  for column in df.columns:
    nunique_values = df[column].nunique()
    print(f"\nNumber of Unique Values of {column}:", nunique_values)

print_nunique_values(df)

Number of Unique Values of User_ID: 5891

Number of Unique Values of Product_ID: 3631

Number of Unique Values of Gender: 2

Number of Unique Values of Age: 7

Number of Unique Values of Occupation: 21

Number of Unique Values of City_Category: 3

Number of Unique Values of Stay_In_Current_City_Years: 5

Number of Unique Values of Marital_Status: 2

Number of Unique Values of Product_Category: 20

Number of Unique Values of Purchase: 18105

Data Cleaning & Preprocessing¶

In [10]:
# change column value in Stay_In_Current_City_Years since it has "+" symbols involved.
df['Stay_In_Current_City_Years'] = df['Stay_In_Current_City_Years'].str.replace('+', '')
In [11]:
# check again
df['Stay_In_Current_City_Years'].unique()
Out[11]:
array(['2', '4', '3', '1', '0'], dtype=object)
In [12]:
# convert to numeric value
df['Stay_In_Current_City_Years'] = pd.to_numeric(df['Stay_In_Current_City_Years'])

Statistical Summary¶

In [13]:
# select numerical columns to see statistics
df.describe(include='all')
Out[13]:
User_ID Product_ID Gender Age Occupation City_Category Stay_In_Current_City_Years Marital_Status Product_Category Purchase
count 5.500680e+05 550068 550068 550068 550068.000000 550068 550068.000000 550068.000000 550068.000000 550068.000000
unique NaN 3631 2 7 NaN 3 NaN NaN NaN NaN
top NaN P00265242 M 26-35 NaN B NaN NaN NaN NaN
freq NaN 1880 414259 219587 NaN 231173 NaN NaN NaN NaN
mean 1.003029e+06 NaN NaN NaN 8.076707 NaN 1.858418 0.409653 5.404270 9263.968713
std 1.727592e+03 NaN NaN NaN 6.522660 NaN 1.289443 0.491770 3.936211 5023.065394
min 1.000001e+06 NaN NaN NaN 0.000000 NaN 0.000000 0.000000 1.000000 12.000000
25% 1.001516e+06 NaN NaN NaN 2.000000 NaN 1.000000 0.000000 1.000000 5823.000000
50% 1.003077e+06 NaN NaN NaN 7.000000 NaN 2.000000 0.000000 5.000000 8047.000000
75% 1.004478e+06 NaN NaN NaN 14.000000 NaN 3.000000 1.000000 8.000000 12054.000000
max 1.006040e+06 NaN NaN NaN 20.000000 NaN 4.000000 1.000000 20.000000 23961.000000
In [14]:
# check for skewness
df.select_dtypes(include = ['int64', 'float64']).skew()
Out[14]:
User_ID                       0.003066
Occupation                    0.400140
Stay_In_Current_City_Years    0.317236
Marital_Status                0.367437
Product_Category              1.025735
Purchase                      0.600140
dtype: float64

Key Observations

  • No missing values in the data
  • Age group 26-35 has the most purchases (219587)
  • Male is more than female (414259)
  • City category B has the most purchases (231173)
  • Min purchase is \$12
  • Max purchase is \$23961

Missing Value Detection¶

In [15]:
df.isnull().sum()
Out[15]:
User_ID                       0
Product_ID                    0
Gender                        0
Age                           0
Occupation                    0
City_Category                 0
Stay_In_Current_City_Years    0
Marital_Status                0
Product_Category              0
Purchase                      0
dtype: int64

Duplicate Value Detection¶

In [16]:
# check for dups
df.duplicated().sum()
Out[16]:
0

Data Visualizations¶

Numerical Features¶

In [17]:
[col for col in df.columns if df[col].dtype in ['int64', 'float64']]
Out[17]:
['User_ID',
 'Occupation',
 'Stay_In_Current_City_Years',
 'Marital_Status',
 'Product_Category',
 'Purchase']
  • Here we'll remove User_ID & Product_Category for our analysis since they're irrelevant.
In [18]:
fig, axis = plt.subplots(nrows = 2, ncols = 2, figsize = (12, 10))
fig.subplots_adjust(top = 0.8)

# Plot distribution plots for specific columns
sns.distplot(df['Occupation'], kde = True, ax = axis[0, 0], color = '#CD5C5C')
sns.distplot(df['Stay_In_Current_City_Years'], kde = True, ax = axis[0, 1], color = '#CD5C5C')
sns.distplot(df['Marital_Status'], kde = True, ax = axis[1, 0], color = '#CD5C5C')

# Plot distribuition plot for `Purchase` with normal curve fit
sns.distplot(df['Purchase'], kde = True, ax = axis[1, 1], color = '#CD5C5C')

# Fitting target variable to the normal curve
mu, sigma = norm.fit(df['Purchase'])
print(f"Mean: {mu}, Standard Deviation: {sigma}")

# Adding legend to the 'Purchase' distribution plot
axis[1, 1].legend(['Normal Distribution (μ = {:.2f}, σ = {:.2f})'.format(mu, sigma)], loc = 'best')

plt.show()

Mean: 9263.968712959126, Standard Deviation: 5023.060827959928
In [19]:
# Count plots for each feature
import plotly.graph_objects as go
from plotly.subplots import make_subplots

# create subplots
fig = make_subplots(rows = 4, cols = 2,
                    subplot_titles = ('Gender', 'Age', 'Occupation', 'City_Category', 'Stay_In_Current_City_Years', 'Marital_Status', 'Product_Category', 'Purchase')
)

# add histograms for each subplot
fig.add_trace(go.Histogram(x = df['Gender'], name = 'Gender'), row = 1, col = 1)
fig.add_trace(go.Histogram(x = df['Age'], name = 'Age'), row = 1, col = 2)
fig.add_trace(go.Histogram(x = df['Occupation'], name = 'Occupation'), row = 2, col = 1)
fig.add_trace(go.Histogram(x = df['City_Category'], name = 'City_Category'), row = 2, col = 2)
fig.add_trace(go.Histogram(x = df['Stay_In_Current_City_Years']), row=3, col=1)
fig.add_trace(go.Histogram(x = df['Marital_Status']), row=3, col=2)
fig.add_trace(go.Histogram(x = df['Product_Category']), row=4, col=1)
fig.add_trace(go.Histogram(x = df['Purchase']), row=4, col=2)

# adjust layout
fig.update_layout(height = 1000, width = 1200, showlegend = False, title_text = 'Count Plots for Each Feature')

fig.show()

Key Obersavations

  • Male : Female = 4 : 1
  • 7 age groups to categorize customers.
  • Most buyers are single.
  • Occupation 8 has the lowest count of customers, can may be ignored.
  • Majority of the products are in 1, 5, and 8. The lowest number of categories can may be combined into a single category to reduce the complexity of the problem.
  • City B has a higher count, indicating more population, following by C and A.
  • Most buyers only live 1 year in the city while the other categories remain in a uniform distribution.

Categorical Features¶

In [20]:
# plot box plots for categorical features
fig, axis = plt.subplots(nrows = 2, ncols = 2, figsize = (12, 10))
fig.subplots_adjust(top = 0.8)

# plotting
sns.boxplot(x = 'Occupation', data = df, ax = axis[0, 0], color = '#CD5C5C')
sns.boxplot(x = 'Stay_In_Current_City_Years', data = df, ax = axis[0, 1], color = '#CD5C5C')
sns.boxplot(x = 'Purchase', data = df, ax = axis[1, 0], color = '#CD5C5C')

# remove any empty plot
plt.tight_layout()

plt.show()

Purchase vs Other Features¶

In [21]:
feat = ['Gender', 'Age', 'Occupation', 'City_Category', 'Stay_In_Current_City_Years', 'Marital_Status', 'Product_Category']

fig, axis = plt.subplots(nrows = 3, ncols = 2, figsize = (12, 16))
fig.subplots_adjust(top = 1.2)

# plotting
count = 0
for i in range(3):
  for j in range(2):
    sns.boxplot(x = feat[count], y = 'Purchase', data = df, ax = axis[i, j], color = '#CD5C5C')
    count += 1
plt.show()

# plot the last feature
plt.figure(figsize = (12, 6))
sns.boxplot(x = feat[-1], y = 'Purchase', data = df, color = '#CD5C5C')
plt.show()
In [22]:
fig, axes = plt.subplots(nrows = 2, ncols = 2, figsize = (20, 6))

fig.subplots_adjust(top = 1.2)

# plot
sns.boxplot(data = df, y = 'Purchase', x = 'Gender', hue = 'Age', ax = axes[0, 0], palette = 'RdBu')
sns.boxplot(data = df, y = 'Purchase', x = 'Gender', hue = 'City_Category', ax = axes[0,1 ], palette = 'RdBu')
sns.boxplot(data = df, y = 'Purchase', x = 'Gender', hue = 'Marital_Status', ax = axes[1, 0], palette = 'RdBu')
sns.boxplot(data = df, y = 'Purchase', x = 'Gender', hue = 'Stay_In_Current_City_Years', ax = axes[1, 1], palette = 'RdBu')

plt.show()
In [23]:
df
Out[23]:
User_ID Product_ID Gender Age Occupation City_Category Stay_In_Current_City_Years Marital_Status Product_Category Purchase
0 1000001 P00069042 F 0-17 10 A 2 0 3 8370
1 1000001 P00248942 F 0-17 10 A 2 0 1 15200
2 1000001 P00087842 F 0-17 10 A 2 0 12 1422
3 1000001 P00085442 F 0-17 10 A 2 0 12 1057
4 1000002 P00285442 M 55+ 16 C 4 0 8 7969
... ... ... ... ... ... ... ... ... ... ...
550063 1006033 P00372445 M 51-55 13 B 1 1 20 368
550064 1006035 P00375436 F 26-35 1 C 3 0 20 371
550065 1006036 P00375436 F 26-35 15 B 4 1 20 137
550066 1006038 P00375436 F 55+ 1 C 2 0 20 365
550067 1006039 P00371644 F 46-50 0 B 4 1 20 490

550068 rows × 10 columns

Data Analysis¶

1. Are women spending more money per transaction than men? Why or why not?

In [24]:
# total purchase amount per customer
df_purchase = df.groupby(['User_ID', 'Gender'])['Purchase'].sum().reset_index()
In [25]:
df_purchase
Out[25]:
User_ID Gender Purchase
0 1000001 F 334093
1 1000002 M 810472
2 1000003 M 341635
3 1000004 M 206468
4 1000005 M 821001
... ... ... ...
5886 1006036 F 4116058
5887 1006037 F 1119538
5888 1006038 F 90034
5889 1006039 F 590319
5890 1006040 M 1653299

5891 rows × 3 columns

In [26]:
# histogram of average amount spent for each customer - Male
df_purchase[df_purchase['Gender'] == 'M']['Purchase'].hist(bins = 100)
plt.show()
In [27]:
# histogram of average amount spent for each customer - Female
df_purchase[df_purchase['Gender'] == 'F']['Purchase'].hist(bins = 100)
plt.show()
In [28]:
male_mean = df_purchase[df_purchase['Gender'] == 'M']['Purchase'].mean()
female_mean = df_purchase[df_purchase['Gender'] == 'F']['Purchase'].mean()

print(f'Average amount spend by Male customers:{male_mean: .2f}' )
print(f'Average amount spend by Female customers: {female_mean: .2f}')

Average amount spend by Male customers: 925344.40
Average amount spend by Female customers:  712024.39
  • Both female and male purchase amount are highly right skewed - mostly people spent less amount of money and only only a few had large purchases. Male customers had an overall high spending trend (over $600) vs female.

2. Confidence intervals and distribution of the mean of the expenses by female and male customers

Bootstrap Method to Estimate CI Based on Gender¶

Why Bootstrap Over Z-Test/T-Test?

  • The bootstrap method is useful when you cannot assume a normal distribution or when the sample size is relatively small.
  • It does not require assumptions about the population distribution, making it more flexible in cases where the underlying data is skewed, as might be the case here with the purchase data.
  • Additionally, since the data may not follow a normal distribution, a Z-test or t-test might not provide reliable confidence intervals, which is why the bootstrap method was chosen.
In [29]:
# 1. separate the data based on gender
male_purchase = df_purchase[df_purchase['Gender'] == 'M']['Purchase']
female_purchase = df_purchase[df_purchase['Gender'] == 'F']['Purchase']
In [30]:
# 2. bootstrap sampling

# set parameters for bootstrap
bootstrap_iterations = 1000
male_sample_size = len(male_purchase)
female_sample_size = len(female_purchase)

# lists to store bootstrap means
bootstrap_male_means = []
bootstrap_female_means = []

# bootstrap process for males and females
for _ in range(bootstrap_iterations):
  male_sample_mean = male_purchase.sample(male_sample_size, replace = True).mean()
  female_sample_mean = female_purchase.sample(female_sample_size, replace = True).mean()

  bootstrap_male_means.append(male_sample_mean)
  bootstrap_female_means.append(female_sample_mean)
In [31]:
# 3. calculate 95% CI from the bootstrap distributions (percentiles)
male_ci = np.percentile(bootstrap_male_means, [2.5, 97.5])
female_ci = np.percentile(bootstrap_female_means, [2.5, 97.5])

print(f'95% Confidence Interval for Male: {male_ci}')
print(f'95% Confidence Interval for Female: {female_ci}')

95% Confidence Interval for Male: [896788.151      955301.36327811]
95% Confidence Interval for Female: [673718.9247599  752440.22519508]
In [32]:
# plot male and female bootstrap distribution
fig, axis = plt.subplots(nrows = 1, ncols = 2, figsize = (20, 6))

axis[0].hist(bootstrap_male_means, bins = 100, color = '#CD5C5C')
axis[1].hist(bootstrap_female_means, bins = 100, color = '#CD5C5C')

axis[0].set_title('Bootstrap Distribution for Male')
axis[1].set_title('Bootstrap Distribution for Female')

plt.show()

3. Are confidence intervals of average male and female spending overlapping? How can Walmart leverage this conclusion to make changes or improvements?

  • The confidence intervals of the two groups are not overlapping.
  • Walmart could introduce gender-specific marketing campaigns. For example, targeting females with promotions that encourage higher spending or focusing on product where females tend to spend more. For men, Walmart can highlight premium products or bundles since the data indicates that they are more likely to make larger purcahses.

4. Results when the same activity is performed for Married vs Unmarried

Bootstrap Based on Marital Status¶

In [33]:
# 1. separate the data based on marital status
married_purchase = df[df['Marital_Status'] == 1]['Purchase']
unmarried_purchase = df[df['Marital_Status'] == 0]['Purchase']
In [34]:
# 2. bootstrap method
bootstrap_iterations = 1000
married_sample_size = len(married_purchase)
unmarried_sample_size = len(unmarried_purchase)

# lists to store bootstrap
bootstrap_married_means = []
bootstrap_unmarried_means = []

# bootstrap
for _ in range(bootstrap_iterations):
  married_sample_mean = married_purchase.sample(married_sample_size, replace = True).mean()
  unmarried_sample_mean = unmarried_purchase.sample(unmarried_sample_size, replace = True).mean()

  bootstrap_married_means.append(married_sample_mean)
  bootstrap_unmarried_means.append(unmarried_sample_mean)
In [35]:
# 3. calculate 95% CI for unmarried and married
ci_unmarried = np.percentile(bootstrap_unmarried_means, [2.5, 97.5])
ci_married = np.percentile(bootstrap_married_means, [2.5, 97.5])

print(f'95% Confidence Interval for Unmarried: {ci_unmarried}')
print(f'95% Confidence Interval for Married: {ci_married}')

95% Confidence Interval for Unmarried: [9248.4936151  9283.62324101]
95% Confidence Interval for Married: [9241.25466845 9283.49212169]
In [36]:
# plot unmarried and married bootstrap distribution
fig, axis = plt.subplots(nrows = 1, ncols = 2, figsize = (20, 6))

axis[0].hist(bootstrap_married_means, bins = 100, color = '#CD5C5C')
axis[1].hist(bootstrap_unmarried_means, bins = 100, color = '#CD5C5C')

axis[0].set_title('Bootstrap Distribution for Married')
axis[1].set_title('Bootstrap Distribution for Unmarried')

plt.show()

Bootstrap on Age Groups¶

5. Results when the same activity is performed for Age

We will analyze the spending patterns of customers in different age groups. By calculating confidence intervals for each age group, we can see which age range tends to spend more and how certain age groups differ in their spending habits.

In [37]:
# 1. list out the age groups in the dataset
age_groups = df['Age'].unique()
In [38]:
age_groups
Out[38]:
array(['0-17', '55+', '26-35', '46-50', '51-55', '36-45', '18-25'],
      dtype=object)
In [39]:
# 2. create dictionary to store bootstrap means for each age group
bootstrap_age_means = {}
In [40]:
# 3. bootstrap iterations
bootstrap_iterations = 1000
In [41]:
# 4. loop over each age group and calculate bootstrap means
for age in age_groups:

  # filter data for the current age group
  age_group_purchase = df[df['Age'] == age]['Purchase']

  # store bootstrap means for this age group
  bootstrap_age_means[age] = []

  for _ in range(bootstrap_iterations):
    sample_mean = age_group_purchase.sample(len(age_group_purchase), replace = True).mean()
    bootstrap_age_means[age].append(sample_mean)
In [42]:
# Now calculate the 95% confidence interval for each age group
ci_age_groups = {}

for age_group, means in bootstrap_age_means.items():
    ci_lower = np.percentile(means, 2.5)
    ci_upper = np.percentile(means, 97.5)
    ci_age_groups[age_group] = (ci_lower, ci_upper)

# Display the confidence intervals for each age group
for age_group, ci in ci_age_groups.items():
    print(f'95% Confidence Interval for Age Group {age_group}: {ci}')

95% Confidence Interval for Age Group 0-17: (8852.336074692093, 9015.506295523772)
95% Confidence Interval for Age Group 55+: (9268.499118768601, 9404.237992931548)
95% Confidence Interval for Age Group 26-35: (9231.860682667917, 9274.961448879032)
95% Confidence Interval for Age Group 46-50: (9164.85631824249, 9255.049930526684)
95% Confidence Interval for Age Group 51-55: (9481.348506532298, 9584.59839874289)
95% Confidence Interval for Age Group 36-45: (9301.65613041186, 9361.617953332789)
95% Confidence Interval for Age Group 18-25: (9139.305498444714, 9201.856779048765)
In [43]:
df.groupby('Age')['Purchase'].sum()
Out[43]:
Age
0-17      134913183
18-25     913848675
26-35    2031770578
36-45    1026569884
46-50     420843403
51-55     367099644
55+       200767375
Name: Purchase, dtype: int64

Final Insights¶

Actionable Insights¶

  • For Age feature, we observed that most customers (over 80%) who belong to the age group 25-40 spend the most
  • For Gender feature, male vs female is around 4:1. Also, male spend more than female:
    • Average amount spend by Male customers: 9,25,408.28
    • Average amount spend by Female customers: 7,12,217.18
  • When combining Purchase with Gender and Marital_Status, we can see that single men spend the most during Black Friday. In other words, it also tells men tend to spend less once they get married.
  • After analyzing Stay_In_Current_Years, we came to know that those who have stayed for 4 years tend to have a wider range of purchase amounts. This also suggests that purchase behavior does not differ dramatically between males and females, but the number of years spent in the current city might have some impact on purchase variability, with longer stays (e.g., 4 years) leading to a higher spread in spending.
  • When examining City_Category, which was the city the product was purchased, even though city B is majorly responsible for the overall sales income, on average, purchases in City C are higher. This suggests that while more transactions occur in City B, the individual transaction values might be smaller, compared to those in City C, which has fewer customers but higher average spending per customer.
  • For Product_Category, 6 and 15 stand out as high-value categories with considerable variability, meaning they likely contain more expensive products. Categories 4, 12, 13, 19, and 20 involve lower purchase amounts and show less variability, which suggests these might be more affordable or consistent-price products. However, product 1, 5, 8, 11 have highest purchasing frequency.

Confidence Interval¶

  1. Gender:
  • 95% Confidence Interval for Male: [895805.64118935 954106.02928402]
  • 95% Confidence Interval for Female: [670968.74483794 750643.14072629]
  1. Marital Status:
  • 95% Confidence Interval for Unmarried: [9249.15868534 9284.20717963]
  • 95% Confidence Interval for Married: [9240.17078742 9281.06980312]
  1. Age group:
  • 95% Confidence Interval for Age Group 0-17: (8860.55, 9017.56)
  • 95% Confidence Interval for Age Group 55+: (9269.97, 9400.03)
  • 95% Confidence Interval for Age Group 26-35: (9232.11, 9273.29)
  • 95% Confidence Interval for Age Group 46-50: (9165.11, 9254.88)
  • 95% Confidence Interval for Age Group 51-55: (9481.92, 9580.95)
  • 95% Confidence Interval for Age Group 36-45: (9301.16, 9361.39)
  • 95% Confidence Interval for Age Group 18-25: (9140.45, 9198.78)

Recommendations¶

  1. Since males have a higher confidence interval for purchase amounts compared to females, consider focusing high-value product marketing or promotional campaigns towards male customers, as they seem to spend more. For Marital Status-Based Offers:

  2. There is not a significant difference between the purchase behaviors of unmarried and married individuals. Therefore, marital status should not be the primary segmentation criterion for differentiated marketing strategies.

  3. Age Groups 51-55 and 55+ have higher upper limits in their confidence intervals, indicating that these age groups tend to make higher purchases. Consider offering premium or higher-end products to these demographics. For younger age groups like 18-25 and 26-35, while their spending is relatively consistent, focusing on products that appeal to price-conscious consumers with moderate spending might be more effective.